Nine times out of ten, CPFind found nothing because the overlap between your photos contains nothing it can recognise — too little overlap, a blank sky or water surface, or wrong lens data pulled from missing EXIF. Add overlap, check the lens type and field of view, and if the automatic pass still fails, add a handful of control points by hand: Hugin only needs a few good ones per image pair to align. This guide works through every cause in order. More Hugin fixes live on the troubleshooting hub.
What “no control points” actually means
Control points are pairs of matching features — a window corner, a rooftop edge, a distinctive rock — that appear in two overlapping photos. Hugin’s optimiser uses them to work out where each photo was taken from and how the lens distorted it. When you click Align in the Assistant, Hugin runs CPFind, its built-in control point detector, over every overlapping pair. If CPFind cannot find enough trustworthy matches between a pair, that pair is left unconnected — and if too many pairs fail, the Assistant stops and reports that it could not align the images.
The important thing to understand is that this is almost never a software fault. CPFind is good at its job, but it can only match what is actually there. Work through the four causes below — they account for virtually every report of this error.
Cause 1: Not enough overlap between frames
CPFind needs a healthy band of shared content between neighbouring photos. If you shot with only 10–15% overlap, the shared strip is narrow, often distorted at the frame edges, and may contain too few distinct features. Hugin’s own documentation and long-standing community guidance is to shoot with roughly 25–50% overlap — err on the generous side, especially with wide-angle lenses whose edges are heavily distorted.
You cannot add overlap after the fact, but you can check whether it is your problem: open the two photos side by side and look at the shared region. If the common area is a thin sliver, that pair will always struggle. For an existing set, manual control points (below) are the rescue path; for the next shoot, follow the overlap and technique rules in our guide to shooting panoramas for Hugin.
Cause 2: The overlap is featureless
Even generous overlap fails when the shared area has no texture: clear blue sky, calm water, fog, snowfields, sand, or a plain white wall. CPFind matches local contrast patterns, and a gradient of sky offers none. Classic cases are a row shot across a lake (all overlap is water and sky), or an interior where the overlap is an empty ceiling.
Fixes: include some textured content in each frame when you shoot — tilt down slightly so rooftops, treelines or foreground objects enter the overlap. For an existing set, place manual control points on whatever structure does exist at the edges of the overlap, and use vertical line points on buildings (covered below). Points do not have to sit in the middle of the overlap; anywhere both photos can see the same real-world feature works.
Cause 3: Wrong lens data because EXIF is missing
CPFind and the optimiser both depend on knowing your lens: its type (normal/rectilinear, fisheye, and so on) and its horizontal field of view (HFOV). Hugin normally calculates these from the EXIF data in your files — focal length and sensor size. If the EXIF has been stripped (common with some phone exports, messaging-app copies, scanned images, or aggressive editing pipelines), Hugin has to guess, and a badly wrong HFOV makes real features land in unexpected places, so matching fails or produces nonsense.
Check in the Photos tab: select an image and look at the lens data Hugin assigned. If the focal length or crop factor is blank or obviously wrong, enter the correct values manually — the real focal length of the lens and your camera’s crop factor are enough for Hugin to compute the HFOV. If you shot with a fisheye, make sure the lens type says so; telling Hugin a fisheye is a normal rectilinear lens is a guaranteed alignment failure.
Cause 4: CPFind needs a second, guided pass
Sometimes the first automatic pass simply leaves gaps: one pair in a long row failed while its neighbours succeeded. Before adding points by hand, try re-running detection with the pair selected. In the Photos tab you can select the images and run Create control points (CPFind) again — a fresh pass sometimes picks up matches the Assistant’s batch run skipped, particularly after you have corrected the lens data in Cause 3. Always fix lens data first; re-running CPFind on wrong data just reproduces the failure.
The reliable fix: add control points manually
Manual points sound laborious, but most failed pairs need only three to six well-placed points to join the project. Open the Control Points tab, choose the first image on the left and its neighbour on the right, and work like this:
- Click a sharp, distinctive feature in the left image — a corner, a sign letter, a window frame. Zoom in for precision.
- Click the same feature in the right image. Hugin refines the second click with a local correlation search; check the magnified view and nudge if needed.
- Click Add to commit the point.
- Spread points around the overlap — corners and middle, near and far — rather than clustering them in one spot.
- Repeat until the pair has at least three points, then move to the next unconnected pair.
Avoid points on moving objects (people, cars, waves, branches in wind) and on clouds: they were somewhere else when the second photo was taken, and they will poison the optimisation later — see our guide to optimizer errors for how bad points show up in the numbers. After adding points, run the optimiser and check the preview before stitching.
Use line points on architecture
Straight structures give you a second kind of point. In the Control Points tab you can add vertical or horizontal line points: place two points along the same real-world edge — the side of a building, a door frame, the horizon — in each image. These tell Hugin “this edge is truly vertical”, which both levels the panorama and helps alignment in scenes where corners are scarce but edges are plentiful. A couple of vertical lines per pair makes city and interior panoramas dramatically better behaved.
Prevent it on the next shoot
Nearly every control point failure is baked in at capture time. Shoot 25–50% overlap, lock exposure and focus, keep some texture in every overlap, and — if your camera or phone strips EXIF on export — export full-size originals with metadata intact. The full checklist is in our panorama shooting guide. If alignment now works but the stitch itself fails at the blending stage, that is a different problem: see the Enblend errors guide.
FAQ
How many control points does Hugin need between two photos?
Three is the practical minimum for a pair, and they must be spread out, not clustered. CPFind usually generates dozens per pair; when you add points by hand, three to six careful ones per failed pair are normally enough for the optimiser to place the images.
Why does CPFind fail on my sky and sea panoramas?
Because there is nothing to match. CPFind looks for local texture and contrast; clear sky, calm water, fog and snow have almost none. Include textured content — rooftops, treeline, foreground — in the overlapping area, or add manual points on the features that do exist.
Hugin says my images are not connected. Is that the same problem?
Yes, it is the other side of it. An image with no control points to any neighbour is ‘not connected’ and cannot be positioned. Add manual points between it and an adjacent image, or re-run CPFind after fixing the lens data.
Can I stitch photos with no EXIF data at all?
Yes, but you must supply the lens data yourself: the lens type and either the focal length plus crop factor, or the horizontal field of view directly. Without believable lens data, both CPFind and the optimiser work blind.
